[Sidefx-houdini-list] [Mantra 12.x] command line -V2+a crashes randomly Mantra on CentOS?
szymon.kapeniak at gmail.com
Sun Jan 27 19:41:15 EST 2013
2013/1/28 Pablo Giménez Pizarro <pablogipi at gmail.com>
> This is strange we have been using Mantra with CentOS for the last 6
> months, and we haven't notice something like this.
> Is this happening with all your scenes?
Yes, many scenes, many builds, in a couple of last months, but it wasn't
easy to track down, because setting verbosity on Mantra ROP works fine. It
has to be overwritten with command like -V option AND beyond level 2a, to
make a crash. Even then crashes happen in 10% cases or less. I don't trust
my self here, the reason might be somewhere else, but I'm out of ideas.
Our 4 last jobs were touched by this decease (we had to re-render holes in
The only reason I've finally yet accidentally discovered it was that I was
modifying our renderfarm ROP in an ongoing improvement process, and one of
the changes was removing command line -V overwrite from a script sent to
farm at all. This cured mantra immediately. I was rendering whole Friday
different sequences from three different projects: with a new setup I
haven't lost a single frame, adding -V5a to a command line, and voila, my
renders are as Swiss cheese.
> Enviado desde mi iPhone
> El 25 Jan 2013, a las 22:04, Szymon Kapeniak <szymon.kapeniak at gmail.com>
> > Hi everyone,
> > Have any one here noticed this strange behavior of Mantra 12. and 12.1:
> > our CentOS 6.2 it crashes in about 5-10% cases at the very start of the
> > rendering just after Mantra launch:
> > stdout:
> > mantra Version 12.1.185 (Compiled on 01/09/13")
> > .. and crash...
> > - this happens only for verbosity level above 2[a] specified by -V
> > - this doesn't happen with "ray_property renderer verbose 5" inside ifd
> > file.
> > - this happens in different builds across 12.0 and 12.1 including newest
> > production build 185
> > I'm testing only production scenes with lots of deep rasters (17+), but
> > many textures nor geometries. It's completely random, 3 to 10 frames per
> > 117 long sequence always on different machines. It also seems like -V2a
> > works just fine, but -V3a crashes... weird...
> > It might be helpful if any someone else would confirm it. It seems like a
> > scene has to be a real one, with some stuff inside ifds, not just sphere,
> > but I'm not sure of that... also, as I said it's just 5-10% of frames.
> > thanks,
> > skk.
> > _______________________________________________
> > Sidefx-houdini-list mailing list
> > Sidefx-houdini-list at sidefx.com
> > https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
> Sidefx-houdini-list mailing list
> Sidefx-houdini-list at sidefx.com
More information about the Sidefx-houdini-list